Mercurial > projects > dwt-addons
annotate dwtx/jface/internal/text/html/HTML2TextReader.d @ 158:25f1f92fa3df
...
| author | Frank Benoit <benoit@tionex.de> |
|---|---|
| date | Tue, 26 Aug 2008 02:46:34 +0200 |
| parents | f70d9508c95c |
| children | 1a5b8f8129df |
| rev | line source |
|---|---|
| 129 | 1 /******************************************************************************* |
| 2 * Copyright (c) 2000, 2008 IBM Corporation and others. | |
| 3 * All rights reserved. This program and the accompanying materials | |
| 4 * are made available under the terms of the Eclipse Public License v1.0 | |
| 5 * which accompanies this distribution, and is available at | |
| 6 * http://www.eclipse.org/legal/epl-v10.html | |
| 7 * | |
| 8 * Contributors: | |
| 9 * IBM Corporation - initial API and implementation | |
| 10 * Port to the D programming language: | |
| 11 * Frank Benoit <benoit@tionex.de> | |
| 12 *******************************************************************************/ | |
| 13 module dwtx.jface.internal.text.html.HTML2TextReader; | |
| 14 | |
| 131 | 15 import dwtx.jface.internal.text.html.HTMLPrinter; // packageimport |
| 16 import dwtx.jface.internal.text.html.BrowserInformationControl; // packageimport | |
| 17 import dwtx.jface.internal.text.html.SubstitutionTextReader; // packageimport | |
| 18 import dwtx.jface.internal.text.html.HTMLTextPresenter; // packageimport | |
| 19 import dwtx.jface.internal.text.html.BrowserInput; // packageimport | |
| 20 import dwtx.jface.internal.text.html.SingleCharReader; // packageimport | |
| 21 import dwtx.jface.internal.text.html.BrowserInformationControlInput; // packageimport | |
| 22 import dwtx.jface.internal.text.html.HTMLMessages; // packageimport | |
| 23 | |
| 129 | 24 import dwt.dwthelper.utils; |
| 158 | 25 import dwtx.dwtxhelper.PushbackReader; |
|
153
f70d9508c95c
Fix java Collection imports
Frank Benoit <benoit@tionex.de>
parents:
150
diff
changeset
|
26 import dwtx.dwtxhelper.Collection; |
|
f70d9508c95c
Fix java Collection imports
Frank Benoit <benoit@tionex.de>
parents:
150
diff
changeset
|
27 |
| 129 | 28 import dwt.DWT; |
| 29 import dwt.custom.StyleRange; | |
| 30 import dwtx.jface.text.TextPresentation; | |
| 31 | |
| 32 | |
| 33 /** | |
| 34 * Reads the text contents from a reader of HTML contents and translates | |
| 35 * the tags or cut them out. | |
| 36 * <p> | |
| 37 * Moved into this package from <code>dwtx.jface.internal.text.revisions</code>.</p> | |
| 38 */ | |
| 39 public class HTML2TextReader : SubstitutionTextReader { | |
| 40 | |
| 147 | 41 private static const String EMPTY_STRING= ""; //$NON-NLS-1$ |
| 42 private static const Map fgEntityLookup; | |
| 43 private static const Set fgTags; | |
| 129 | 44 |
| 150 | 45 static this() { |
| 129 | 46 |
| 47 fgTags= new HashSet(); | |
| 48 fgTags.add("b"); //$NON-NLS-1$ | |
| 49 fgTags.add("br"); //$NON-NLS-1$ | |
| 50 fgTags.add("br/"); //$NON-NLS-1$ | |
| 51 fgTags.add("div"); //$NON-NLS-1$ | |
| 52 fgTags.add("h1"); //$NON-NLS-1$ | |
| 53 fgTags.add("h2"); //$NON-NLS-1$ | |
| 54 fgTags.add("h3"); //$NON-NLS-1$ | |
| 55 fgTags.add("h4"); //$NON-NLS-1$ | |
| 56 fgTags.add("h5"); //$NON-NLS-1$ | |
| 57 fgTags.add("p"); //$NON-NLS-1$ | |
| 58 fgTags.add("dl"); //$NON-NLS-1$ | |
| 59 fgTags.add("dt"); //$NON-NLS-1$ | |
| 60 fgTags.add("dd"); //$NON-NLS-1$ | |
| 61 fgTags.add("li"); //$NON-NLS-1$ | |
| 62 fgTags.add("ul"); //$NON-NLS-1$ | |
| 63 fgTags.add("pre"); //$NON-NLS-1$ | |
| 64 fgTags.add("head"); //$NON-NLS-1$ | |
| 65 | |
| 66 fgEntityLookup= new HashMap(7); | |
| 67 fgEntityLookup.put("lt", "<"); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 68 fgEntityLookup.put("gt", ">"); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 69 fgEntityLookup.put("nbsp", " "); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 70 fgEntityLookup.put("amp", "&"); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 71 fgEntityLookup.put("circ", "^"); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 72 fgEntityLookup.put("tilde", "~"); //$NON-NLS-2$ //$NON-NLS-1$ | |
| 73 fgEntityLookup.put("quot", "\""); //$NON-NLS-1$ //$NON-NLS-2$ | |
| 74 } | |
| 75 | |
| 76 private int fCounter= 0; | |
| 77 private TextPresentation fTextPresentation; | |
| 78 private int fBold= 0; | |
| 79 private int fStartOffset= -1; | |
| 80 private bool fInParagraph= false; | |
| 81 private bool fIsPreformattedText= false; | |
| 82 private bool fIgnore= false; | |
| 83 private bool fHeaderDetected= false; | |
| 84 | |
| 85 /** | |
| 86 * Transforms the HTML text from the reader to formatted text. | |
| 87 * | |
| 88 * @param reader the reader | |
| 89 * @param presentation If not <code>null</code>, formattings will be applied to | |
| 90 * the presentation. | |
| 91 */ | |
|
133
7d818bd32d63
Fix ctors to this with gvim regexp
Frank Benoit <benoit@tionex.de>
parents:
131
diff
changeset
|
92 public this(Reader reader, TextPresentation presentation) { |
| 129 | 93 super(new PushbackReader(reader)); |
| 94 fTextPresentation= presentation; | |
| 95 } | |
| 96 | |
|
136
6dcb0baaa031
Regex removal of throws decls, some instanceof
Frank Benoit <benoit@tionex.de>
parents:
134
diff
changeset
|
97 public int read() { |
| 129 | 98 int c= super.read(); |
| 99 if (c !is -1) | |
| 100 ++ fCounter; | |
| 101 return c; | |
| 102 } | |
| 103 | |
| 104 protected void startBold() { | |
| 105 if (fBold is 0) | |
| 106 fStartOffset= fCounter; | |
| 107 ++ fBold; | |
| 108 } | |
| 109 | |
| 110 protected void startPreformattedText() { | |
| 111 fIsPreformattedText= true; | |
| 112 setSkipWhitespace(false); | |
| 113 } | |
| 114 | |
| 115 protected void stopPreformattedText() { | |
| 116 fIsPreformattedText= false; | |
| 117 setSkipWhitespace(true); | |
| 118 } | |
| 119 | |
| 120 protected void stopBold() { | |
| 121 -- fBold; | |
| 122 if (fBold is 0) { | |
| 123 if (fTextPresentation !is null) { | |
| 124 fTextPresentation.addStyleRange(new StyleRange(fStartOffset, fCounter - fStartOffset, null, null, DWT.BOLD)); | |
| 125 } | |
| 126 fStartOffset= -1; | |
| 127 } | |
| 128 } | |
| 129 | |
| 130 /* | |
| 131 * @see dwtx.jdt.internal.ui.text.SubstitutionTextReader#computeSubstitution(int) | |
| 132 */ | |
|
136
6dcb0baaa031
Regex removal of throws decls, some instanceof
Frank Benoit <benoit@tionex.de>
parents:
134
diff
changeset
|
133 protected String computeSubstitution(int c) { |
| 129 | 134 |
| 135 if (c is '<') | |
| 136 return processHTMLTag(); | |
| 137 else if (fIgnore) | |
| 138 return EMPTY_STRING; | |
| 139 else if (c is '&') | |
| 140 return processEntity(); | |
| 141 else if (fIsPreformattedText) | |
| 142 return processPreformattedText(c); | |
| 143 | |
| 144 return null; | |
| 145 } | |
| 146 | |
| 147 private String html2Text(String html) { | |
| 148 | |
| 149 if (html is null || html.length() is 0) | |
| 150 return EMPTY_STRING; | |
| 151 | |
| 152 html= html.toLowerCase(); | |
| 150 | 153 |
| 129 | 154 String tag= html; |
| 155 if ('/' is tag.charAt(0)) | |
| 156 tag= tag.substring(1); | |
| 157 | |
| 158 if (!fgTags.contains(tag)) | |
| 159 return EMPTY_STRING; | |
| 160 | |
| 161 | |
| 162 if ("pre".equals(html)) { //$NON-NLS-1$ | |
| 163 startPreformattedText(); | |
| 164 return EMPTY_STRING; | |
| 165 } | |
| 166 | |
| 167 if ("/pre".equals(html)) { //$NON-NLS-1$ | |
| 168 stopPreformattedText(); | |
| 169 return EMPTY_STRING; | |
| 170 } | |
| 171 | |
| 172 if (fIsPreformattedText) | |
| 173 return EMPTY_STRING; | |
| 174 | |
| 175 if ("b".equals(html)) { //$NON-NLS-1$ | |
| 176 startBold(); | |
| 177 return EMPTY_STRING; | |
| 178 } | |
| 179 | |
| 180 if ((html.length() > 1 && html.charAt(0) is 'h' && Character.isDigit(html.charAt(1))) || "dt".equals(html)) { //$NON-NLS-1$ | |
| 181 startBold(); | |
| 182 return EMPTY_STRING; | |
| 183 } | |
| 184 | |
| 185 if ("dl".equals(html)) //$NON-NLS-1$ | |
| 186 return LINE_DELIM; | |
| 187 | |
| 188 if ("dd".equals(html)) //$NON-NLS-1$ | |
| 189 return "\t"; //$NON-NLS-1$ | |
| 190 | |
| 191 if ("li".equals(html)) //$NON-NLS-1$ | |
| 192 // FIXME: this hard-coded prefix does not work for RTL languages, see https://bugs.eclipse.org/bugs/show_bug.cgi?id=91682 | |
| 193 return LINE_DELIM + HTMLMessages.getString("HTML2TextReader.listItemPrefix"); //$NON-NLS-1$ | |
| 194 | |
| 195 if ("/b".equals(html)) { //$NON-NLS-1$ | |
| 196 stopBold(); | |
| 197 return EMPTY_STRING; | |
| 198 } | |
| 199 | |
| 200 if ("p".equals(html)) { //$NON-NLS-1$ | |
| 201 fInParagraph= true; | |
| 202 return LINE_DELIM; | |
| 203 } | |
| 204 | |
| 205 if ("br".equals(html) || "br/".equals(html) || "div".equals(html)) //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ | |
| 206 return LINE_DELIM; | |
| 207 | |
| 208 if ("/p".equals(html)) { //$NON-NLS-1$ | |
| 209 bool inParagraph= fInParagraph; | |
| 210 fInParagraph= false; | |
| 211 return inParagraph ? EMPTY_STRING : LINE_DELIM; | |
| 212 } | |
| 213 | |
| 214 if ((html.startsWith("/h") && html.length() > 2 && Character.isDigit(html.charAt(2))) || "/dt".equals(html)) { //$NON-NLS-1$ //$NON-NLS-2$ | |
| 215 stopBold(); | |
| 216 return LINE_DELIM; | |
| 217 } | |
| 218 | |
| 219 if ("/dd".equals(html)) //$NON-NLS-1$ | |
| 220 return LINE_DELIM; | |
| 150 | 221 |
| 129 | 222 if ("head".equals(html) && !fHeaderDetected) { //$NON-NLS-1$ |
| 223 fHeaderDetected= true; | |
| 224 fIgnore= true; | |
| 225 return EMPTY_STRING; | |
| 226 } | |
| 150 | 227 |
| 129 | 228 if ("/head".equals(html) && fHeaderDetected && fIgnore) { //$NON-NLS-1$ |
| 229 fIgnore= false; | |
| 230 return EMPTY_STRING; | |
| 231 } | |
| 232 | |
| 233 return EMPTY_STRING; | |
| 234 } | |
| 235 | |
| 236 /* | |
| 237 * A '<' has been read. Process a html tag | |
| 238 */ | |
|
136
6dcb0baaa031
Regex removal of throws decls, some instanceof
Frank Benoit <benoit@tionex.de>
parents:
134
diff
changeset
|
239 private String processHTMLTag() { |
| 129 | 240 |
| 241 StringBuffer buf= new StringBuffer(); | |
| 242 int ch; | |
| 243 do { | |
| 244 | |
| 245 ch= nextChar(); | |
| 246 | |
| 247 while (ch !is -1 && ch !is '>') { | |
| 134 | 248 buf.append(Character.toLowerCase(cast(wchar) ch)); |
| 129 | 249 ch= nextChar(); |
| 250 if (ch is '"'){ | |
| 134 | 251 buf.append(Character.toLowerCase(cast(wchar) ch)); |
| 129 | 252 ch= nextChar(); |
| 253 while (ch !is -1 && ch !is '"'){ | |
| 134 | 254 buf.append(Character.toLowerCase(cast(wchar) ch)); |
| 129 | 255 ch= nextChar(); |
| 256 } | |
| 257 } | |
| 258 if (ch is '<' && !isInComment(buf)) { | |
| 259 unread(ch); | |
| 260 return '<' + buf.toString(); | |
| 261 } | |
| 262 } | |
| 263 | |
| 264 if (ch is -1) | |
| 265 return null; | |
| 266 | |
| 267 if (!isInComment(buf) || isCommentEnd(buf)) { | |
| 268 break; | |
| 269 } | |
| 270 // unfinished comment | |
| 134 | 271 buf.append(cast(wchar) ch); |
| 129 | 272 } while (true); |
| 273 | |
| 274 return html2Text(buf.toString()); | |
| 275 } | |
| 150 | 276 |
| 129 | 277 private static bool isInComment(StringBuffer buf) { |
| 278 return buf.length() >= 3 && "!--".equals(buf.substring(0, 3)); //$NON-NLS-1$ | |
| 279 } | |
| 150 | 280 |
| 129 | 281 private static bool isCommentEnd(StringBuffer buf) { |
| 282 int tagLen= buf.length(); | |
| 283 return tagLen >= 5 && "--".equals(buf.substring(tagLen - 2)); //$NON-NLS-1$ | |
| 284 } | |
| 285 | |
| 286 private String processPreformattedText(int c) { | |
| 287 if (c is '\r' || c is '\n') | |
| 288 fCounter++; | |
| 289 return null; | |
| 290 } | |
| 291 | |
| 292 | |
|
136
6dcb0baaa031
Regex removal of throws decls, some instanceof
Frank Benoit <benoit@tionex.de>
parents:
134
diff
changeset
|
293 private void unread(int ch) { |
| 134 | 294 (cast(PushbackReader) getReader()).unread(ch); |
| 129 | 295 } |
| 296 | |
| 297 protected String entity2Text(String symbol) { | |
| 298 if (symbol.length() > 1 && symbol.charAt(0) is '#') { | |
| 299 int ch; | |
| 300 try { | |
| 301 if (symbol.charAt(1) is 'x') { | |
| 302 ch= Integer.parseInt(symbol.substring(2), 16); | |
| 303 } else { | |
| 304 ch= Integer.parseInt(symbol.substring(1), 10); | |
| 305 } | |
| 134 | 306 return EMPTY_STRING + cast(wchar)ch; |
| 129 | 307 } catch (NumberFormatException e) { |
| 308 } | |
| 309 } else { | |
| 134 | 310 String str= cast(String) fgEntityLookup.get(symbol); |
| 129 | 311 if (str !is null) { |
| 312 return str; | |
| 313 } | |
| 314 } | |
| 315 return "&" + symbol; // not found //$NON-NLS-1$ | |
| 316 } | |
| 317 | |
| 318 /* | |
| 319 * A '&' has been read. Process a entity | |
| 320 */ | |
|
136
6dcb0baaa031
Regex removal of throws decls, some instanceof
Frank Benoit <benoit@tionex.de>
parents:
134
diff
changeset
|
321 private String processEntity() { |
| 129 | 322 StringBuffer buf= new StringBuffer(); |
| 323 int ch= nextChar(); | |
| 134 | 324 while (Character.isLetterOrDigit(cast(wchar)ch) || ch is '#') { |
| 325 buf.append(cast(wchar) ch); | |
| 129 | 326 ch= nextChar(); |
| 327 } | |
| 328 | |
| 329 if (ch is ';') | |
| 330 return entity2Text(buf.toString()); | |
| 331 | |
| 332 buf.insert(0, '&'); | |
| 333 if (ch !is -1) | |
| 134 | 334 buf.append(cast(wchar) ch); |
| 129 | 335 return buf.toString(); |
| 336 } | |
| 337 } |